RU2828154C1 - Breathing pattern monitoring device (embodiments), method of operating breathing pattern monitoring device (embodiments), wearable device including breathing pattern monitoring device - Google Patents

Abstract

FIELD: medicine.

SUBSTANCE: group of inventions relates to medicine, namely to methods for controlling breathing patterns and wearable devices for realizing said methods. Personal data of the user are entered into the memory unit. User PPG signal is received from photoplethysmographic sensor. Data is received from the sensor of the inertial measuring unit IMU. Respiratory signal and heart rate (HR) signal are separated from PPG signal in separation unit. Type of physical activity of the user is recognized by the data from the IMU sensor and the data of the heart rate signal in the unit for recognizing the type of physical activity. Actual breathing pattern is separated from the respiratory signal and parameters of the actual breathing pattern in the classification unit. Parameters of the actual breathing pattern are compared in the comparison unit with the parameters of the optimal breathing pattern from the database recommended for the recognized physical activity. Deviations of the parameters of the actual breathing pattern of the user from the parameters of the optimal breathing pattern from the database are detected in the comparison unit. Using the unit for issuing recommendations to the user on the breathing technique based on the detected deviations of the parameters of the actual breathing pattern of the user from the parameters of the optimal breathing pattern from the database.

EFFECT: provides for independent control of breathing technique by users and/or development of correct respiratory habits based on monitoring of their breathing by means of portable devices.

13 cl, 7 dwg

Description

Field of technology

The present invention relates to monitoring human breathing, namely to monitoring breathing patterns using a wearable device.

Description of the Prior Art

Currently, smart devices worn on the user's body, such as smart watches, fitness trackers, headphones, etc., as well as personal medical devices, are widely used to monitor human health. Based on the information received from sensors located in such devices, the user or the attending physician receives health information and recommendations for improving well-being.

Among all human biophysical processes, breathing is unique because, in addition to the purely biochemical and instinctive nature of breathing, a person can consciously control the breathing process. A large number of scientific studies indicate that the quality of everyday activities (e.g., sports, meditation, rest, walking, etc.) can be significantly improved by consciously monitoring and managing breathing. In addition, conscious breathing practices can improve mental activity, have a calming effect on a person under stress, improve the quality of sleep, and increase the body's endurance. That is, continuous monitoring of breathing both in the waking state and in sleep can help the user identify and consciously correct the influence of critical factors on the user's condition.

Breathing patterns reflect the formation of the body's adaptation mechanisms during physical activity. For high performance in sports, optimal functioning of the external respiratory system is important [1, 2, 3]. One cycle of alternating inhalation and exhalation constitutes the respiratory act. Usually, inhalation is somewhat shorter than exhalation; in humans, their ratio is approximately 1:3. The body selects a breathing pattern in which the required level of alveolar ventilation is achieved with the least expenditure of energy on the work of the respiratory muscles. Compliance with this condition, together with maintaining the balance of inhaled and exhaled air, ensures an optimal breathing pattern - an optimal ratio of the components of the respiratory act (duration of phases, depth of breathing, dynamics of pressure and flows in the airways). The normal volume of air inhaled in one breath is 500 ml, and at rest - about 5-8 liters of air per minute [4, 5].

Fig. 1 shows spirograms of pathological types of breathing and normal breathing. Violation of the optimal pattern and rhythm of breathing can reflect a pathological process in the body, leaving its mark on the processes of breathing regulation. When the optimal pattern of breathing is disrupted, breathing becomes shallow and is mainly carried out by the neck muscles. There are several types of pathological breathing, as shown in Fig. 1. The most common pathologies of breathing include those associated with a violation of the balance between excitation and inhibition in the central nervous system. Pathological (periodic) breathing is external breathing, which is characterized by a group rhythm, often alternating with pauses (periods of breathing alternate with periods of apnea) or with intercalated periodic inhalations.

Examples of respiratory pathologies are the following.

Bradypnea is a rare breathing that occurs when the respiratory center is damaged and depressed due to hypoxia, edema, ischemia, and the effects of narcotic substances. When large airways are narrowed, rare and deep breathing is observed - stenotic. An increase in blood pressure causes a reflex decrease in the frequency of respiratory movements.

Tachypnea is frequent shallow breathing. It occurs due to pronounced stimulation of the respiratory center during hypercapnia, hypoxemia. This type of breathing is observed during fever, pneumonia, pulmonary congestion, atelectasis, etc.

Hyperpnea is deep and frequent breathing, observed during muscular work, hypercapnia, emotional stress, thyrotoxicosis, anemia, acidosis, and decreased oxygen content in the inhaled air.

Apnea - the absence of respiratory movements is observed with hypocapnia, decreased excitability of the respiratory center due to damage to the brain, exo- and endointoxication, and the effects of narcotic substances.

Cheyne-Stokes breathing is characterized by alternating groups of respiratory movements with increasing amplitude and periods of apnea (breathing arrest). It should be noted that Cheyne-Stokes breathing can occur not only in pathological conditions, but also in the norm: in healthy people during sleep, in premature babies with an immature respiratory regulation system, in old age (when there is an increase in the threshold of excitability of the respiratory center).

Kussmaul breathing is a large, noisy, deep breathing ("breathing of a cornered animal") characterized by separate convulsive contractions of the main and accessory respiratory muscles. This type of breathing indicates deep hypoxia of the brain. It is characteristic of uremic and diabetic coma and terminal conditions of other genesis.

Biot's respiration is characterized by alternating periods of apnea with groups of respiratory movements of equal amplitude.

Atactic breathing is uneven, chaotic, irregular breathing. It is observed when the respiratory neurons of the medulla oblongata are preserved, but when the connection with the respiratory neurons of the pons is disrupted.

It is well known that breathing patterns can be measured, classified and quantified. Provided that breathing patterns are monitored 24 hours a day, it is possible to monitor the user's condition and improve their physical condition through individual breathing recommendations. The prior art includes devices used to determine optimal breathing patterns, such as thermistors, impedance pneumography, and spirometers [6, 7]. However, these methods cause discomfort to the patient during measurement, so the proposed method for determining the optimal breathing pattern using photoplethysmography is preferable [5, 8].

In terms of physiology, monitoring breathing patterns would help reduce the risks of deterioration of a person's physical condition, such as the functioning of the cardiovascular system, the brain, the sensation of chronic pain, etc. With long-term monitoring of breathing patterns, recommendations excluding certain exercises would help the user improve his health, as well as recommended individual exercises that help improve the quality of breathing, and warnings to the user about the danger to the body associated with improper breathing, for example, during sleep, would also help.

Mentally, long-term monitoring of breathing patterns would help combat stress by recommending specific breathing techniques or exercises during stress detected by monitoring.

If the quality of breathing deteriorates due to the environment in which the user is located, it is possible to give individual recommendations, for example, how to minimize health risks in a particular unfavorable environmental situation.

Continuous monitoring of breathing quality during exercise would allow us to provide recommendations on breathing practices suitable for performing a particular physical exercise.

Currently, methods and devices are known for monitoring breathing patterns at any stage of the breathing process, but the known devices are not portable wearable devices and are not capable of monitoring the breathing of an active and practically healthy person for a long time. In the known devices, only the breathing rate is extracted from the signal.

The physiological and emotional state of a person, as well as the intensity of the activity, determine the limits within which a person can consciously or unconsciously choose the actual breathing pattern. Without conscious control, a person chooses a “default” pattern, which is determined by the principle of “minimum effort” for the current metabolic needs. However, the “default” pattern may not be optimal for a particular type of human activity, for example, in the case of sports, the performance and benefit from the exercise can be significantly improved by conscious control of breathing.

The article "Breath Tools: A Synthesis of Evidence-Based Breathing Strategies to Enhance Human Running", Front. Physiol., 17 March 2022, Sec. Exercise Physiology, Volume 13-2022, https://doi.org/10.3389/fphys.2022.813243, examines the parameters of breathing pattern as a "breathing tool" that can be consciously used to achieve athletic goals. Diaphragmatic breathing is reported to be correlated with various positive health benefits, such as reduced resting heart rate, post-exercise oxidative stress, increased postural control, and baroreflex sensitivity. Nasal breathing during exercise can result in decreased respiratory rate (RR), decreased hypocapnia, and increased nitric oxide production. Nitric oxide itself is beneficial as a vasodilator and bronchodilator. It has been shown that nasal breathing reduces the respiratory exchange ratio, the maximum volume of oxygen ( _VO2 ) that can be received by the body, transferred to the muscles and used to generate energy in the aerobic mode. Nasal breathing also increases the efficiency of running and the time to fatigue. Breathing with long exhalations (33% versus 50% constant air flow) causes an improvement in heart rate variability, an increase in the efficiency of lung ventilation and _VO2 . Active exhalations cause a more complete upward excursion of the diaphragm, which reduces the work of the diaphragm and promotes postural stabilization.

A system and method for extracting breathing patterns from a photoplethysmogram (PPG signals) document US 11389118 B2, publication date 07/19/2022 are known from the prior art. The document discloses a filter by which the PPG signal is filtered from signals received from a user's wearable device. The filtered PPG signal indicates the user's breathing pattern. The known solution does not evaluate breathing patterns, nor does it disclose information on how to use breathing patterns; in addition, there is no parametric evaluation of breathing patterns.

Also known from the prior art is a wearable pulse oximeter and breathing monitor (document US 20220175249 A1, publication date 06/09/2022). The known device for monitoring a patient's condition is small and light enough to be comfortably worn on a patient, for example, on the patient's wrist or around the neck. The patient monitoring device may have a monitoring tool that is removably attached to a disposable base. The base may have outputs for connecting to an acoustic breathing sensor and a pulse oximeter sensor. The patient monitoring device may have connectors with movable pins connecting the monitor device and the disposable stand so that the monitor device can receive data from sensors connected to the disposable stand. The known solution does not evaluate breathing patterns, and the user does not receive any recommendations on how to use breathing patterns.

Also known from the prior art is a plethysmographic determination of the respiratory rate (document US 20190076028 A1, publication date 03/14/2019). The known device is a plethysmographic respiratory processor that responds to respiratory effects appearing on the blood volume curve and the corresponding detected intensity waveform measured by an optical sensor in a section of peripheral tissue perfused with blood to ensure the measurement of the respiratory rate. The processor identifies a portion of the plethysmogram corresponding to a physiologically acceptable series of pulses of the plethysmograph signal. Several processors output various parameters depending on the specific respiratory effects on the plethysmogram. The decision logic determines the respiratory rate based on at least some of these parameters. The disadvantages of the known solution are that only the respiratory rate is determined, there is no assessment of the breathing pattern, no individual recommendations are given to the user regarding breathing, and the known device is not portable.

A biological information analysis device is known from the prior art (document EP 3427653 B1, publication date 16.01.2019). In the known solution, one breathing pattern from a plurality of breathing patterns is evaluated, its effect on blood pressure is assessed, and feedback is provided when the breathing pattern changes to change blood pressure. The disadvantages of the known solution are that the feedback is limited only to recommendations regarding the breathing pattern for changes in blood pressure, the solution is only used in ambulatory devices and is not used in wearable portable devices.

Currently, there are no easy-to-use portable devices in the state of the art that could help users independently control their breathing technique and/or develop correct breathing habits based on monitoring their breathing. Therefore, a solution is needed that would rely on existing sensors (PPG, accelerometers, etc.), would be used in portable wearable devices and would allow for quantitative determination of the user's breathing pattern and provide clear recommendations for improving breathing technique for a particular activity.

The essence of the invention

A system for monitoring breathing patterns using a wearable device is proposed.

A breathing pattern monitoring device is proposed, which contains:

a memory unit designed to store the user's personal data, as well as a database storing optimal breathing patterns recommended for each type of physical activity based on the user's personal data,

a photoplethysmographic (PPG) sensor located on the user's body, configured to receive the user's PPG signal;

an inertial measurement unit (IMU) sensor located on the user's body;

a separation unit configured to extract from the PPG signal:

- respiratory signal and

- heart rate signal;

a physical activity type recognition unit configured to recognize the user's physical activity type based on IMU sensor data and heart rate signal data;

a classification unit configured to extract an actual breathing pattern from a respiratory signal and to recognize breathing parameters of the actual breathing pattern;

a comparison unit configured to compare the parameters of the actual breathing pattern from the database with the parameters of the optimal breathing pattern recommended for the recognized physical activity, extracted from the database, and also configured to detect a deviation of the parameters of the user's actual breathing pattern from the parameters of the optimal breathing pattern from the database recommended for the recognized physical activity;

a recommendation issuing unit configured to issue recommendations to the user on breathing technique based on the detected deviations of the parameters of the user's actual breathing pattern from the parameters of the optimal breathing pattern from the database recommended for the recognized physical activity.

Moreover, the separation block may contain spectral filters. Personal data may include height, weight, age, gender. Breathing parameters are: volume of inhaled air, inhalation rate, exhalation time, volume of exhaled air, respiratory rate, time between exhalation and subsequent inhalation, duration of the breathing cycle, depth of breathing, duration of breath holding. Recommendations may be displayed on the screen of the user's wearable device.

A method for monitoring breathing patterns using the proposed breathing pattern monitoring device is proposed, comprising the steps of:

enter the user's personal data into the memory block;

receive the user's PPG (photoplethysmogram) signal from the photoplethysmographic (PPG) sensor;

receive data from the inertial measurement unit (IMU) sensor;

the respiratory signal and the heart rate signal are separated from the PPG signal in a separation unit;

recognize the type of physical activity of the user based on data from the IMU sensor and heart rate signal data in the physical activity type recognition unit;

extract the actual breathing pattern from the respiratory signal and the parameters of the actual breathing pattern in the classification block;

compare in the comparison block the parameters of the actual breathing pattern with the parameters of the optimal breathing pattern from the database recommended for the recognized physical activity;

identify in the comparison block deviations of the parameters of the user's actual breathing pattern from the parameters of the optimal breathing pattern from the database recommended for the recognized physical activity;

issue, via the issuing unit, recommendations to the user on breathing technique based on the detected deviations of the parameters of the user's actual breathing pattern from the parameters of the optimal breathing pattern from the database recommended for the recognized physical activity.

Moreover, the identification of the actual breathing pattern is carried out by performing the following steps:

for each respiratory parameter extracted from the respiratory signal:

a) arbitrarily set the duration of the time window,

b) calculate the average value each breathing parameter over a given time window;

c) calculate the standard deviation from the average value ;

d) compare the standard deviation with an average value ;

If for at least 70% of the breathing parameters, the breathing pattern is analyzed with a given time window duration,

If for at least 70% of the breathing parameters, then reduce the duration of the time window and perform steps b-d,

If for at least 70% of the breathing parameters, then increase the duration of the time window and perform steps b-d.

After entering personal data, it is additionally possible to perform a calibration measurement to identify respiratory pathology, and if respiratory pathology is identified, the user receives a recommendation to contact a doctor to compile an individual database.

Also offered is a breathing pattern monitoring device containing:

a memory unit designed to store the user's personal data and a trained neural network;

a photoplethysmographic (PPG) sensor located on the user's body, configured to receive the user's PPG signal;

an inertial measurement unit (IMU) sensor located on the user's body;

a separation unit configured to separate a respiratory signal from the PPG signal;

a trained neural network designed to generate an optimal respiratory signal based on the user's personal data, respiratory signal, and signal from the IMU sensor;

a recommendation issuing unit configured to issue recommendations to the user on breathing technique based on the optimal respiratory signal issued by the neural network.

Moreover, personal data includes height, weight, age, gender. The neural network can be additionally implemented with the ability to determine whether the user is a practically healthy person or has a breathing pattern related to a pathological breathing pattern; if it is determined that the breathing pattern relates to a pathological breathing pattern, the issuing unit issues a recommendation to consult a doctor to compile an individual database of optimal breathing patterns corresponding to various types of physical activity.

A method for monitoring breathing patterns using a breathing pattern monitoring device is also provided, comprising the steps of:

enter the user's personal data into the memory block;

receive the user's PPG (photoplethysmogram) signal from the photoplethysmographic (PPG) sensor;

receive data from the inertial measurement unit (IMU) sensor;

extract the respiratory signal from the PPG signal by means of a separation unit,

the user's personal data, the mentioned respiratory signal and the signal from the IMU sensor are fed to the input of the trained neural network;

based on the input data, an optimal respiratory signal is obtained at the output of the neural network;

issue, via the issuing unit, a recommendation to the user on breathing technique based on the optimal respiratory signal issued by the neural network.

Variants of wearable devices are proposed, including variants of devices for monitoring breathing patterns, designed with the possibility of implementing the corresponding method for monitoring breathing patterns.

Brief description of the drawings

The above and other features and advantages of the present invention are explained in the following description, illustrated by the drawings, which show the following:

Fig. 1 illustrates examples of spirograms of pathological types of breathing and a normal spirogram.

Fig. 2 illustrates examples of phase portraits of possible breathing patterns and their acceptable values in a system of two breathing parameters: depth and frequency of breathing, A) for a healthy person, B) for a person in a weak physical condition, C) for a person under stress.

Fig. 3 illustrates a respiratory signal describing the breathing process, which is an example of a breathing pattern.

Fig. 4 illustrates a diagram of the proposed method for monitoring breathing patterns.

Fig. 5 illustrates two breathing patterns: (A), (C) - optimal breathing pattern, (B) - the user's actual measured breathing pattern.

Fig. 6 illustrates the steps for determining the deviation of the user's breathing pattern from the optimal breathing pattern from the database.

Fig. 7 illustrates the experimental part.

Detailed description of the invention

The proposed invention allows monitoring the quality of the user's breathing patterns using a portable wearable device, on the basis of which conclusions are made about the user's physical condition in real time and about the breathing technique required at the current moment in time to improve the physical condition. Thanks to the proposed invention, the user can use the wearable device to control the breathing technique to improve his physical condition at the current moment in time.

The PPG signal (photoplethysmogram) contains oscillations corresponding to the heart rate, which are modulated by fluctuations in respiratory activity, i.e. it contains a respiration-induced modulation (respiratory signal). The respiratory signal can be extracted by measuring PPG signals using known techniques using a wearable device. The respiratory signal is usually considered as a parasitic signal and is removed during PPG signal processing.

In the present invention, the respiratory signal is filtered from the PPG signal and used equally with the heart rate signal. For this purpose, the original PPG signal is filtered using two filters. One filter (high-pass filter) extracts from the original PPG signal the part of the signal corresponding to the heart rate. The main difference of the filter for extracting the heart rate is in the frequency band that needs to be preserved, for heart contractions this is approximately from 0.5 Hz to 1.5 Hz. It is possible to use a band-pass filter configured to pass the signal in the specified band, and it is also possible to use a higher-pass filter. The second filter extracts from the original PPG signal the part of the signal corresponding to the respiratory modulation (respiratory signal).

The part of the PPG signal corresponding to the heart rate is responsible for the type of physical activity of the user, and the part of the PPG signal corresponding to the respiratory modulation (respiratory signal) is responsible for the shape of the respiratory signal, i.e. the dependence of the signal amplitude on time. Unlike the solutions known from the prior art, the proposed invention involves selecting and storing the entire frequency band occupied by the respiratory signal and storing all the harmonics present in the respiratory signal. It should be noted that in the respiratory signal, by analogy with the sound signal, it is possible to select the main frequency and harmonics at frequencies multiple of the main frequency. It is the frequency of the harmonics, the amplitude and the phase that contain information about the shape of the respiratory signal. The spectrum of the respiratory signal can be complex and can contain a large number of harmonics. As mentioned above, a spectral filter is used to select the spectrum of the respiratory signal, wherein the amplitude, frequency and phase characteristics of the filter are selected in such a way as to transmit information (amplitude, frequency, phase) about the harmonics of the respiratory signal with minimal losses. The passband of such a filter is wider than the frequency band of the respiratory signal, and the shape of the filter passband does not contain significant oscillations in the region of the lower and upper cutoff frequencies (oscillations at the edges of the filter passband). The presence of such oscillations at the edges of the filter passband in the AFC (amplitude frequency response) or PFC (phase frequency response) of the filter means that the signal frequencies lying in the region of the lower and upper cutoff frequencies of the filter are transmitted with amplitude or phase distortions, which, in turn, leads to distortion of the respiratory signal after passing the filter. The absence of oscillations at the edges of the filter passband should lead to the preservation of the shape of the respiratory signal and allows extracting much more information about the user's breathing.

A wearable device is any compact user device, such as a smart watch, ring, earphone, clip, etc., that allows for a long period of time to non-invasively and without user intervention record a photoplethysmographic signal and transmit the recorded signal or the result of intermediate processing of the recorded signal to a host device (e.g. a smartphone) for subsequent processing, or perform the processing independently with subsequent issuance of the result of the breathing pattern analysis and recommendations for its correction to the user.

A breathing pattern is a set of breathing parameters extracted from a respiratory signal over a certain period of time (time window), the combination of which is the only possible way to characterize the features of the breathing process of a certain user under certain conditions. Breathing parameters may include (but are not limited to) the following: breathing rate, breathing depth, inhalation rate (time), exhalation rate (time), inhaled air volume, exhaled air volume, breath-holding time, number and frequency of breath-holds, the average value of the listed parameters over a certain period of time, deviations of the listed parameters from the average values over a certain period of time, etc. A breathing pattern is determined by how inhalations and exhalations look, how they change over a certain period of time (time window). The time window is the period of time during which one pattern lasts, and, accordingly, the number of inhalations and exhalations in it are selected depending on the variability of the breathing parameters, as will be described below. Also, in some embodiments, the duration of the time window may be adjusted during operation if, for example, the breathing pattern changes too quickly or too slowly.

The respiratory signal may remain stable over a long period of time (e.g., during human sleep), i.e., the variability of the respiratory parameters will be low, and each respiratory parameter will be fairly stable throughout the selected time window, i.e., throughout the actual breathing pattern. In such cases, the actual breathing pattern with a long time window duration can be selected. In situations of increased user physical activity, the respiratory signal may be so unstable that the respiratory parameters will vary greatly even over a short period of time, in which case the actual breathing pattern with a short time window duration is extracted from the respiratory signal. Therefore, the time window duration of the actual breathing pattern is adjusted dynamically as the user is monitored, i.e., as the breathing parameters are monitored over time.

Dynamic adjustment of the time window of the actual breathing pattern is performed as follows. First, the initial duration of the time window is arbitrarily selected, for example, 5 acts of breathing, i.e. 5 inhalations-exhalations, corresponding to approximately 20 seconds. For the section of the respiratory signal with the selected duration of the time window, the breathing parameters are calculated, for example, such as those indicated above. Since the breathing parameters can be calculated separately for each act of breathing (each inhalation-exhalation), the average value can be calculated on the section of the respiratory signal with the selected duration of the time window each breathing parameter and standard deviation (root mean square deviation) each breathing parameter from its average value . If for at least 70% of the breathing parameters it turns out to be less , then it is concluded that the variability of the breathing parameters during the selected duration of the time window is low. Therefore, further, based on the calculations and comparisons performed, the duration of the time window is increased, for example, by two times. The averages are calculated again and deviations for the new increased duration of the time window and their comparison is performed. The described operation is performed until the deviation for more than 70% of the breathing parameters will not become equal . In this case, the duration of the time window when the condition was met , and there is that period of time that corresponds to one breathing pattern, which is described by the calculated breathing parameters and their deviations (or dispersion).

If already at the first stage, for the time window of the original duration, for at least 70% of the parametersit will turn out to be more, it is concluded that the variability of breathing parameters with such a time window duration is high and it is necessary to reduce the duration of the time window. The duration of the time window is reduced, for example, by half, the averages are calculated and compared againand standard deviationsparameters of respiration for the new duration of the time window. The described operation is repeated until for more than 70% of the parameters of respiration the standard deviationwill not become equal to the average value .

A situation may arise when it is no longer practical to reduce the window duration, but the condition not yet fulfilled, for example, when the number of breaths in the window is five. In this case, the reduction of the time window duration is stopped and the current values are fixed And . In this way, a set of breathing parameters with high variability and a corresponding breathing pattern are obtained.

When equilibrium was established according to the conditions described above, the current actual breathing pattern is considered in the selected time window. However, it is possible that the determination of the current pattern takes too much time because the condition not completed.

Let us consider, for example, a situation in which the variability of respiratory parameters remains low, , over a long period of time, with an increase in the time window. And vice versa, a situation in which the variability of breathing parameters can be high even with the shortest time window and remain so for a long time. In both cases, the user may never receive an assessment of the current breathing pattern and the corresponding recommendations. In this case, a maximum time interval is set, after which the user is guaranteed to receive an assessment of the current breathing pattern, even if the condition is not met . This time interval is set either automatically, or the user can independently set such an interval, according to his preferences, how often he wants to receive notifications about optimal breathing, for example, once a day during everyday activities, or once every thirty minutes during active physical activity.

Recommendations can be issued for one actual pattern or for a group of actual patterns. In this case, if a person experiences active physical activity, it makes sense to track each actual pattern in order to provide a recommendation for adjustment in time during the load. If a person is engaged in a measured type of activity that does not require active intervention in real time, it is possible to collect statistics consisting of a large number of actual patterns per day, per week, etc. and provide summary information. The most accurate recommendations are issued for a long period of time and when monitoring a group of actual patterns.

To determine the optimal breathing pattern of any user, a predictive model is created in advance as a result of machine learning of the neural network, which allows determining the optimal breathing pattern of any user for a certain type of physical activity. Optimal breathing patterns for various types of physical activity are known and recommended by doctors based on studies of various groups of people, based on the physiological needs of the body for different types of human physical activity.

To train the neural network, a training database is compiled.

The training data set is collected from a diverse population of test participants with a wide range of variations in the physiological parameters of the test participants (height, weight, age, gender, etc.) and with different types of physical activity. The database contains all subgroups important for the study, namely statistically almost healthy test participants and test participants with pathological breathing patterns. The participants are divided into subgroups in a medical institution using medical methods of breathing pattern analysis known from the state of the art. At the beginning of the examination, a spirogram is taken from each of the participants. If the participant's breathing pattern signal on the spirogram has a form corresponding to the normal breathing type, as shown in Fig. 1, then this participant is included in the subgroup of statistically almost healthy participants. If the participant's breathing pattern signal on the spirogram has a form corresponding to a variant of deviation from the norm (one of the types of pathological breathing pattern, as shown in Fig. 1), then this participant is included in the subgroup of participants with pathological breathing.

After dividing the participants into subgroups, data is collected from each participant as follows:

- When the participant is in a resting state, PPG signals and IMU signals are read from the inertial measurement unit (IMU) sensors and from the photolethysmograph built into the wearable device worn by the participant. A respiratory signal is extracted from the PPG signal using signal processing methods. The IMU signal indicates the absence of physical activity. The obtained respiratory signal will correspond to the absence of physical activity. The training data set includes the participant's personal data (e.g., gender, age, weight, height) and the obtained respiratory signal data and IMU signal data corresponding to the personal data, as corresponding to the absence of physical activity.

- Based on the personal data of each participant, the parameters of the optimal breathing pattern for each type of physical activity are determined; such parameters are known from the state of the art. It is known that the optimal breathing pattern is the one in which the required level of alveolar ventilation is achieved with the least energy expenditure on the work of the respiratory muscles during physical exertion in combination with the volume of air inhaled in one breath of 500 ml, and at rest 5-8 liters of air per minute. From here, the optimal breathing parameters of a practically healthy person can be calculated.

Before performing physical activity, the reserves of the respiratory system are determined and, ultimately, the efficiency of the respiratory function as a whole is determined; such methods are known from the state of the art (for example, Kuchkin, S.N. Respiratory system reserves (review and state of the problem) / S.N. Kuchkin // Respiratory system reserves. - Volgograd, 1999. - P. 7-51).

Then the state of the external respiration function of each participant is determined by spirography and pneumotachography. The following parameters are analyzed: vital capacity of the lungs, inspiratory reserve beat, tidal volume, etc. The analysis of the parameters obtained by spirography and pneumotachography is carried out in accordance with the established limits of the norm and the limits of deviation of the parameters of external respiration, as well as in accordance with the limits of normal values and deviations from the norm of the main parameters of the flow-volume curve (in % of the expected value); such parameters are known from the state of the art (see, for example, Turina O.N. Organization of work on studying the functional state of the lungs by spirography and pneumotography methods and application of these methods in clinical practice: Method. Instructions / O.I. Turina et al. - Minsk, 2002, 78 p.). For example, normally the flow-volume curve has an almost triangular shape. The time to reach the peak velocity is reached in the first 0.1 s, then there is a smooth decrease in flow. The duration of forced exhalation should be at least 6 seconds in adults and at least 3 seconds in children.

When the participant performs physical activity, PPG signals and IMU signals are recorded from the inertial measurement unit (IMU) sensors and from the photolethysmograph built into the participant's wearable device. At the same time, the participant's breathing parameters are monitored using methods known from the prior art (e.g., by recording a spirogram). Before the study, the health care professional instructs the user on how the user should optimally breathe, and during the study, the health care professional monitors the readings of the devices monitoring the participant's breathing parameters and instructs the participant on how to change their breathing so that the readings correspond to the readings during optimal breathing.

From the PPG signal, using signal processing methods, a respiratory signal is extracted. The IMU signal indicates the presence of physical activity, the type of physical activity is known. The obtained respiratory signal and IMU signal will correspond to the physical activity that the participant performed, breathing optimally. The training data set includes the participant's personal data (e.g., gender, age, weight, height), the corresponding respiratory signal data, IMU signal data, and the type of physical activity. That is, in this case, the respiratory signal will correspond to the optimal breathing pattern of the participant, since he breathed optimally, which was confirmed by monitoring the breathing parameters during the physical activity.

It should be noted that the determination of the type of physical activity based on the IMU sensor data and heart rate signal data is known from the prior art, for example from the documents: Hnoohom N., Mekruksavanich S., Jitpattanakul A. "Physical Activity Recognition Based on Deep Learning Using Photoplethysmography and Wearable Inertial Sensors", Electronics 2023, 12(3), 693; https://doi.org/10.3390/electronics12030693; Biagetti, G.; Crippa, P.; Falaschetti, L.; Orcioni, S.; Turchetti, C. Human Activity Recognition Using Accelerometer and Photoplethysmographic Signals. In Proceedings of the Intelligent Decision Technologies 2017, Algarve, Portugal, 21-23 June 2017; Czarnowski, I., Howlett, R. J., Jain, L. C., Eds.; Springer International Publishing: Cham, Switzerland, 2018; pp. 53-62.3. Casson, A. J.; Vazquez Galvez, A.; Jarchi, D. Gyroscope vs. accelerometer measurements of motion from wrist PPG during physical exercise. ICT Express 2016, 2, 175-179. [Google Scholar] [CrossRef].

When fed personal data of a specific user and the actual breathing pattern to the input of a trained neural network, the trained neural network can:

determine the type of physical activity of the user,

determine whether the user is a generally healthy person or has a breathing pattern that is related to a pathological breathing pattern;

if it is determined that the user is a practically healthy person, the trained neural network determines the optimal breathing pattern for the user, and also compares the parameters of the user's actual breathing obtained from the user's actual breathing pattern with the breathing parameters obtained from the optimal breathing pattern for a given type of physical activity,

Based on the work of the neural network, the user is given:

if it is determined that the current breathing pattern is pathological, it is recommended to consult a doctor to create an individual database of optimal breathing patterns corresponding to a certain type of physical activity,

if the current breathing pattern is determined to be that of a generally healthy person, a recommendation on the nature of optimal breathing that the user should adhere to at that moment in time.

For simplicity, the measured breathing pattern of a specific user and the acceptable limits of typical average statistical patterns can be visualized using two parameters - breathing depth and breathing rate. Fig. 2 illustrates examples of phase portraits of possible breathing patterns and their acceptable values in the breathing depth and breathing rate system. In Fig. 2, the circles with the largest diameters represent the physiological limits of the norm and outline the areas corresponding to the typical breathing pattern, the typical physiological limits, i.e. the norm. Each circle in Fig. 2 does not represent a function of the dependence of breathing rate on breathing depth, but limits the geometric locus of points corresponding to:

- physiological limits of breathing of the average person (large circle in figure 2 (A), (B), (C)),

- the breathing limits of the average person under stress (the average circle in Figure 2 (A), (B), (C)),

- the breathing limits of the average person during physical activity (the smallest circle in figure 2 (A), (B), (C)).

Any point within these areas represents a certain combination of the depth and frequency of breathing, which fit within the limits of one pattern. That is, Fig. 2 shows phase portraits (a set of possible states of the system, described in the space of its parameters), possible breathing patterns and their permissible values in the system of two breathing parameters: the depth and frequency of breathing A) for a practically healthy person, B) for a person in a weak physical condition, C) for a person under stress. In all three variants:

- the designated point is a breathing pattern arbitrarily chosen by the user during a certain type of physical activity, that is, the user himself arbitrarily and unconsciously chose how to breathe;

- the star marked is the optimal breathing pattern for a given type of physical activity.

The relationship between the depth of breathing and the frequency of breathing for a healthy person shows that a person can use almost all types of breathing within their physical capabilities (large circumference). For current physical activity, a small adjustment in breathing can improve the performance of physical activity. For example, when running, deep diaphragmatic breathing will help to increase the activation of the abdominal muscles and improve the process of muscle formation.

In Fig. 2 (B) the dependence of the depth of breathing on the respiratory rate of a person whose health is weakened shows that the physical limits of breathing for such a person are significantly reduced. The large circle, marked with a solid line, shows the limitation of breathing for a practically healthy person, and the circles of a smaller diameter show the limitation of breathing for a physically weakened person. It is evident that during physical activity the optimal breathing pattern (marked with an asterisk) is not available for a person whose health is weakened. In this case, it is necessary to reduce the level of physical activity or completely abandon physical activity. That is, the physical activity chosen by the user will be either useless or harmful, which means the user should abandon this type of physical activity and select another type of physical activity that is suitable for his physiological capabilities.

Fig. 2 (C) shows an example of a person in a highly stressful situation, in which the optimal breathing parameters for any activity cannot be achieved in principle (asterisk outside the circle). It is clear that the optimal breathing pattern (the pattern shown by the asterisk) cannot be achieved due to the user's high stress. In this case, it is possible to recommend postponing physical activity until the person has calmed down sufficiently, or to recommend meditation or special breathing exercises.

When determining the optimal breathing pattern, the user's stress level data can be taken into account. It is known from the state of the art that modern smart watches and fitness bracelets have a function for measuring the user's stress level. The device analyzes the time difference between successive heartbeats (pulse variability). To obtain more accurate information, pulse variability monitoring occurs continuously in order to better determine what variability between beats occurs during training and during rest. It should be taken into account that the readings during training cannot be used to directly determine the stress level, since stress is an integral component of any physical activity. The smart device determines the stress level when a person is at rest (in the absence of physical activity). The device analyzes the user's breathing (inhalation and exhalation). When inhaling, there is a slight acceleration of the heartbeat and the time difference between beats is reduced, i.e. the sympathetic nervous system is activated. When exhaling, the reverse process occurs - the parasympathetic system is activated, the pulse slows down, and the duration between heartbeats increases. One of the main indicators of stress in the body is a disruption of this process, i.e. very low pulse variability is observed (even if the pulse is very low). For a more accurate calculation of stress, the following physiological characteristics of the user are taken into account, obtained using a photoplethysmographic (PPG) sensor built into the device: pulse, pulse variability, respiratory rate, maximum oxygen consumption (VO ₂ max), and excess oxygen consumption after exercise (EPOC).

If the user's breathing pattern differs from the optimal one, which is recommended for a given type of activity taking into account the physiological characteristics of a particular user, this can lead, for example, to early fatigue, ineffective training when the muscles do not receive enough oxygen, or the necessary muscle groups do not work, as well as serious complications with the heart or other organs, etc. Optimal breathing affects the reduction of fatigue, the efficiency of oxygen supply to the muscles and brain, the possible duration of activity, the time and efficiency of recovery after exercise, etc.

The average optimal breathing pattern can be determined not only using a neural network, as described above, but also extracted from a database. In general, a standard database includes data on gender, age, height and weight, i.e. personal data of an average practically healthy person, and recommended optimal breathing patterns (i.e. sets of breathing parameters) corresponding to personal data for different types of physical activity of an average practically healthy person with corresponding profile data.

The optimal breathing patterns calculated by the trained neural network may differ for the same physical activity for different people and represent a range of acceptable values of the optimal breathing parameters for a certain type of physical activity with known personal data of the user. That is, each set of profile data corresponds to at least one pair of data, including a type of physical activity with the optimal breathing pattern corresponding to this type of physical activity. The standard database is formed on the basis of examining a group of people under controlled conditions as part of clinical studies similar to collecting a training database for training the neural network. That is, the optimal breathing pattern for a certain type of physical activity in the database is the average optimal breathing pattern for the examined participants with approximately the same personal data. The accuracy of issuing recommendations when using the database is lower than the accuracy of issuing recommendations when using the trained neural network.

However, neither the trained neural network nor the standard database take into account the individual characteristics of the user, and are recommended to users who are practically healthy, since they offer optimal breathing patterns for the average person. Thus, the user has a choice - either to use the standard options - the trained neural network or the standard database stored in the memory of the breathing pattern monitoring device, or to undergo an examination and create their own individual database to receive accurate individual recommendations.

In one embodiment, determining the optimal breathing patterns for different types of physical activity for a specific user can be done during a direct visit by the user to a doctor, for example, a pulmonologist, i.e. a specialist who studies and treats diseases of all parts of the respiratory system, considering them in close connection with the whole body. When determining the optimal breathing patterns for a specific user, the doctor takes into account all personal data of the user, such as weight, height, age, blood pressure, heart rate at rest and under load, the presence of chronic diseases, bad habits, as well as all physiological characteristics that can affect the breathing process. In addition, for more accurate predictions of optimal breathing patterns for different types of activity, the user is offered to undergo some additional examinations, such as, for example, an ECG or Holter monitoring. The doctor also measures breathing patterns arbitrarily selected by the user, for different types of activity, with and without loads, as well as changes in breathing patterns in response to recommended actions to correct them. When measuring the user's breathing patterns, the doctor monitors blood pressure, pulse, and changes in the ECG. Based on the results of the examination, the doctor compiles an individual database that includes recommended optimal breathing patterns for various user states, i.e. depending on the amplitude, frequency, and phase of the respiratory wave in the breathing pattern for various types of physical activity. In this case, the user enters data from the database received from the doctor into a breathing pattern monitoring device or into a wearable device that uses a breathing pattern monitoring device. The database contains information about various types of physical activity and the optimal breathing patterns for a given user, corresponding to each type of physical activity. In other words, the optimal sets of respiratory act characteristics for each type of physical activity for a given user.

The wearable device, in which the proposed breathing pattern monitoring device is used, may contain a trained neural network and/or a memory device for storing a database. The wearable device may contain an input device designed to input personal user data (e.g. age, gender, weight, height), as well as data from an individual database recommended by a doctor.

In one embodiment, the screen of the wearable device displays the actual and optimal breathing patterns, for example, in the form of a dependence of the depth of breathing on time, for a given type of physical activity and the corresponding recommendations for achieving optimal values of breathing patterns for a given type of physical activity. The device may also additionally contain a communication module configured to exchange information with a remote server (for example, a doctor) and/or cloud storage (a doctor or user).

Different activities and different health conditions of a person have different limitations and requirements for optimal breathing. Any breathing analysis and recommendations should be considered in the context of a specific activity and a specific person. Within the same activity, different goals may imply different optimal breathing patterns. For example, when running, a person can realize different goals by consciously regulating their breathing. That is, the user adjusts their breathing to what is determined to be optimal for them in a given situation, and the breathing pattern changes. With the proposed invention, breathing patterns are monitored in real time, and the user receives feedback regarding how optimal the breathing pattern they consciously reproduce is.

For example:

- Cardio running involves a speed at which a person does not overload his heart and lungs, and does not overwork his muscles, and distance, the goal is cardio training, in this case breathing should be carried out in an economical mode, that is, have a moderate frequency and depth. The speed of cardio running is usually about 6-8 km / h. This type of running is the safest and is suitable for people of different ages and levels of physical fitness.

- Sprint involves a short distance and high speed (more than 8 km/h), the goal is to reach maximum speed, in this case breathing should be shallow (i.e. breathing in short breaths), fast and highly synchronized with the movement;

- A marathon involves running for endurance, the goal is to reach the finish line, in this case breathing should be economical, slow, with nasal inhalation, which involves economical use of air taken in during inhalation, that is, gradual exhalation of air).

Continuous monitoring of breathing patterns can identify dangerous conditions for sensitive groups of people such as asthmatics, diabetics, people with sleep apnea, cardiovascular diseases, people in comas, etc.

Fig. 3 illustrates an example of a respiratory signal as a breathing pattern of an average, practically healthy person. The breathing pattern is used to determine a set of parameters of the respiratory act. Time is plotted along the X axis, and the amplitude of the optical signal corresponding to different phases of the respiratory process is plotted along the Y axis. As described above, the respiratory signal, i.e. the signal describing the breathing process, is extracted from the PPG signal recorded by the photoplethysmograph of the wearable device. The graph is obtained from the PPG signal by processing that allows preserving the amplitude, frequency, and phase of the PPG signal components corresponding to the respiratory activity. Processing can be performed using an analog filter, a digital filter, machine learning methods, neural networks, etc.

Such a graph (Fig. 3) is very informative, since each part of the graph corresponds to some characteristic of the respiratory act, for example, but without limitations:

1 - volume of inhaled air;

2 - inhalation rate;

3 - exhalation time;

4 - volume of exhaled air;

5 - respiratory rate;

6 - time between exhalation and subsequent inhalation (pause);

7 - duration of the breathing cycle;

8 - depth of breathing;

9 - duration of breath holding;

10 - holding your breath.

The dotted line shows the tangent to the leading edge of the pulse corresponding to the inspiratory phase. The slope of the tangent determines the inspiratory velocity 2.

It is also possible to determine the number and frequency of breath holding, the average value of the listed parameters for a certain period of time, deviations of the listed parameters from the average values for a certain period of time, etc.

It is known from the prior art to determine the respiratory parameters from the respiratory signal, as shown in Fig. 3 (P. H Charlton1, T. Bonnici, L. Tarassenko, J. Alastruey, D. A Clifton, R. Beale1, P. J Watkinson, “Extraction of respiratory signals from the electrocardiogram and photoplethysmogram: technical and physiological determinants”, Physiological Measurement, v. 38 (5), 2017) It should be noted that the list of respiratory parameters is provided only as an example, and this is only one of the possible options for analyzing the respiratory signal.

In Fig. 3, the frame on the right shows the difference between two events - one inhalation/exhalation and another inhalation/exhalation. Graph 11 corresponds to the case when the person inhaled in one way, graph 12 corresponds to the case when the person inhaled in a different way. The differences in the pulse shape of different inhalations show the difference between breathing patterns, as well as the different stability of breathing. If the user's breathing pattern changes during the signal accumulation time, the signal can be divided into sections and each section can correspond to its own pattern. In this way, it is possible to track the dynamics of changes in breathing patterns over an indefinite period of time.

Breathing patterns of the same type may have individual variability for the user due to physiological differences between users. The performance of the algorithms for extracting recommended breathing patterns and classifying/estimating the parameters of actual breathing patterns obtained from the user may be improved if they are calibrated for a specific user. Certain user movement patterns (race walking, running, yoga, meditation, etc.) may be included in the calibration. Before using the proposed breathing pattern monitoring device, the user is recommended to enter personal data, indicating gender, age, height, weight, and also to perform a calibration measurement in a calm state (in the absence of stress and without physical activity) to identify obvious breathing pathologies, such as those shown in Fig. 1. Detection of obvious breathing pathologies is performed by a neural network by comparing a set of breathing parameters obtained from the respiratory signal of the actual current PPG signal with breathing parameters related to pathological breathing types. If the calibration measurement reveals that the user's breathing corresponds to pathological breathing, the user is advised to consult a doctor to compile an individual database, as described above.

When a personalized database compiled by a physician is used, the device operates as described below, but instead of a trained neural network or an average database, a personalized database is used and the data is entered into the wearable device.

Fig. 4 illustrates the scheme of the proposed method for monitoring breathing patterns:

Stage 1 is performed in the wearable device 13, containing a photoplethysmographic (PPG) sensor and an IMU sensor. The user records his personal data (e.g., gender, height, weight, age) in the memory block. Also, at the initial stage, the user can perform a calibration measurement in a calm state to identify obvious breathing pathologies. If no obvious breathing pathologies are detected during the calibration measurement, the photoplethysmographic (PPG) sensor installed in the wearable device produces a PPG signal (photoplethysmogram). The resulting photoplethysmogram can be processed both on the wearable device itself and on a smartphone, on a computer, etc. Data is also received from the IMU sensor.

Stage 2 is performed in the separation unit 14. In the separation unit, the respiratory signal and the heart rate signal are separated from the PPG signal.

In the physical activity type recognition unit 15, the user's physical activity type is recognized based on the heart rate signal and data from the IMU sensor (e.g., sports, sleep, rest, meditation, etc.). In one embodiment of the invention, the user can specify the type of his/her physical activity by entering data on the type of physical activity into the wearable device.

Step 3 is performed in classification block 16. In classification block 16, the actual breathing pattern is recognized from the respiratory signal, i.e. the duration of the time window is set and the parameters (shown in Fig. 3) of the respiratory signal in the selected time window are estimated. If, during the execution of step for at least 70% of the breathing parameters, then the breathing pattern is analyzed with a given time window duration, if for at least 70% of the breathing parameters, then reduce the duration of the time window if for at least 70% of the breathing parameters, then increase the duration of the time window.

Step 4 is performed in the trained neural network. The user's personal data, the actual breathing pattern data, and the IMU sensor data are fed to the trained neural network's input. The neural network outputs a recommendation on the optimal respiratory signal in the form of text or a graph illustrating the optimal respiratory signal superimposed on the actual current respiratory signal.

If a database is used instead of a neural network, then step 4 is performed in comparison unit 17. In comparison unit 17, the parameters of the actual breathing pattern are compared with the parameters of the breathing pattern from the database recommended for the recognized physical activity. Deviations of the characteristics (shown in Fig. 3) of the user's actual breathing pattern from the parameters of the optimal breathing pattern from the database recommended for the recognized physical activity are detected. Then step 5 is performed in recommendation output unit 18. Based on the comparison, a recommendation is issued to the user in real time on the breathing technique that the user can use in this type of physical activity to obtain the best results without harm to their health. That is, if deviations of the parameters of the user's actual breathing pattern from the parameters of the breathing pattern from the database are detected, then the user is issued, as a recommendation, the said breathing pattern from the database. The recommendation can be issued in various ways, for example, in the form of specific verbal recommendations that correct the characteristics of the breathing act shown in Fig. 3. For example, the recommendation may also be displayed in the form of a graph as shown in Fig. 5, as will be described below.

The wearable device can operate in the breathing tracking mode 24/7. When the user selects the standard database, the user must enter into the wearable device or smartphone containing the proposed breathing pattern monitoring device, their profile data, such as gender, height, weight, age, which will be taken into account when issuing optimal recommendations to the user for conscious breathing regulation.

Fig. 5 illustrates two breathing patterns using the example of one characteristic of the respiratory act - breathing depth (i.e., Fig. 5 does not show the entire breathing pattern, which is extracted from the PPG signal, as in Fig. 3, but only one breathing parameter - breathing depth). Such graphs can be constructed for all breathing parameters. The user can switch the device so that the recommended result is displayed in different parameters. (A) optimal breathing depth depending on time, (B) actually measured breathing depth of the user depending on time. In this example, the actual breathing depth (B) is measured over a long period of time (i.e., has a long time window), while the optimal breathing depth (A) (dotted line) and the permissible deviations (C) of breathing depth from the optimal breathing depth are synchronously determined. Such graphs can be displayed on the display of the wearable device in real time. In Fig. 5, the graph shows, as an example, only one of the parameters determining the breathing pattern - breathing depth. The user's task is to literally breathe in such a way that curve (B) coincides with curve (A) (or is within the permissible deviations (C)), for example, by increasing the amplitude of their curve (B), i.e. the depth of breathing, and adjusting the periodicity of breathing, i.e. aligning the period of their curve (B) and the period of curve (A). The user's task is either to independently align the curves if the recommendations are given in the form of graphs. Moreover, with the correct application of the recommendations, graph (A) should coincide with graph (B), or at least the user can achieve a breathing mode in which graph (B) will be within the permissible deviations (C) of breathing from the optimal breathing pattern.

Alternatively, the user can follow verbal guidance that appears on the wearable device and consciously breathe in the recommended manner.

The user can also receive advice in the form of both graphs and verbal recommendations on how to align their graph with the optimal breathing graph.

In this way, the user can control, for example, shortness of breath. That is, when shortness of breath is recognized, the limit of permissible deviations of breathing from the optimal breathing pattern and a recommendation to the user on the breathing method are displayed on the screen. When applying the recommendation, the user can monitor the correctness of the recommendation by the fact that his breathing does not go beyond the limit of permissible deviations of breathing from the optimal breathing pattern. If the duration or intensity of shortness of breath goes beyond the range of optimal values, the user is shown a warning message on adjusting breathing.

In this way, it is possible to monitor various breathing patterns, such as shortness of breath, shallow breathing, normal breathing, deep breathing, etc., and the distribution of these patterns over the observation time, i.e. at what point in time which breathing pattern was observed. Currently, the Galaxy Watch only displays information on sleep quality, stress, heart rate, etc.

To analyze the breathing pattern, it is necessary to detect at least 3 breathing acts, i.e. 3 breaths, the minimum duration of the PPG signal is 5 seconds.

Fig. 6 illustrates the steps for determining the deviation of the user's actual breathing pattern from the optimal breathing pattern, in the case of using a database:

(A) There is the actual breathing pattern of the user, indicated by a point on the phase portrait in the system of two breathing parameters - breathing depth B and breathing frequency A, and the type of physical activity of the user is also determined. The two breathing parameters A and B are shown only as an example, in the general case, it is possible to determine all breathing parameters from the actual breathing pattern, as shown in Fig. 3;

(B) The ellipse indicates the area of breathing patterns from the database that are optimal for the user-specified physical activity;

(C) Determine the smallest distance (deviation) from the actual breathing pattern of the user to the area with breathing patterns from the database that are optimal for the physical activity specified by the user. The deviation ΔA along the x-axis corresponds to the deviation from the optimal (recommended) breathing frequency A, the deviation ΔB along the y-axis corresponds to the deviation from the optimal (recommended) breathing depth B.

(D) Based on the deviations received, the user is given recommendations on how to optimize breathing. The recommendations can be given in any convenient form, for example, in words such as "breathing frequency should be such and such, depth such and such" or "breathe more often/less often, deeper/shallower, through the nose/mouth, chest/stomach". It should be noted that a person cannot breathe with the stomach as such, since there are no lungs in it, it can only be inflated or drawn in. In essence, breathing with the stomach means deep breathing with the diaphragm, which provides the body with a sufficient level of oxygen, as opposed to shallow breathing with the ribcage (chest). Additional studies have shown a clear difference in the shape of the respiratory signal for breathing with the stomach and shallow breathing with the chest.

In another embodiment, the recommendations can be accompanied by visualization, as in Fig. 5, when it is necessary to fit the user's curve (B) into the curve (A) of the recommended breathing, then the user sees in real time how the shape of his curve (B) changes and approaches the recommended curve (A).

In addition to the PPG signal, it is possible to use additional signals received by the wearable device using sensors such as a bioimpedance sensor, gyroscope, microphone, and temperature sensor. In particular, the microphone is used to determine the sound of inhaled/exhaled air, or the sounds of steps and movements to determine the type of physical activity.

If the user considers himself/herself to be practically healthy, but at the same time cannot achieve the recommended breathing pattern within 10-15 minutes, it is considered that this is due to the physiological characteristics of a specific user. In this case, the recommendations are changed, the user is provided with easier conditions and a less optimal breathing pattern suitable for the user's current physical activity and for the user's current physical condition, so that the user has the opportunity to follow the recommendations. Easier conditions are calculated based on the reaction of the user's physiological parameters to the recommendations. If, for example, when increasing the frequency or depth of breathing according to the recommendation, a person's heart rate increases or decreases significantly, then he/she cannot withstand such a proposed pace and other "easier" breathing parameters are recommended. In parallel, the user's profile is updated and new optimal profiles corresponding to his/her condition are selected from the database.

The connection between breathing patterns and the user's physical activity is important, for example, for professional athletes. For example, when doing abdominal exercises from a lying position, it is important for the athlete to exhale at the top point, that is, when the body rises during the exercise, and inhale when the body lowers. When doing abdominal exercises, exhale when raising the legs or body, and inhale when lowering. Breathing should be deep (diaphragmatic), but comfortable. When inhaling, the stomach should inflate. This activates the diaphragm and allows for a much greater effect from the exercise. In this case, the user is recommended a breathing pattern based on the user's specific movement, that is, the user can consciously adjust the breathing parameters to their specific movements, thanks to recommendations received from the wearable device. In the case of exercises, the user may be given comments that they are inhaling too early or too late.

The respiratory signal can be extracted from the PPG signal, for example, by the following methods known from the prior art:

1. Amplitude demodulation.

2. Frequency demodulation.

3. Combination of amplitude and frequency demodulation.

4. Using machine learning algorithms.

5. Adaptive methods.

To inform the user about the relationship between their actual breathing pattern and the optimal breathing pattern, various types of feedback can be used. For example, visual, audio, and vibration feedback. In the most complex case, the device demonstrates the user's target breathing pattern as a time graph of some breathing parameter (for example, breathing depth). For exercises, it is more convenient to use audio or vibration signals synchronized with the target breathing pattern. In this case, the user does not need to immediately look at the wearable device.

Thanks to the proposed invention, a positive effect on the health of sensitive groups of users, such as asthmatics and allergy sufferers, can be achieved by recommending optimal breathing patterns depending on the actual environmental conditions. For example, avoiding deep breaths for people with respiratory diseases during periods of falling air quality index (AQI). In addition, the wearable device can be integrated with a device that determines the chemical composition of inhaled and exhaled air, such devices are widely known, for example, https://metabolicratetest.com/, or integrated with a device capable of calculating the ratio between the volumes of exhaled carbon dioxide and the volume of inhaled oxygen.

Based on the parameters of the actual breathing patterns of the user, it is possible to accurately estimate the total energy expenditure (TEE) during the user's physical activity. The TEE value can be accurately estimated using certain parameters of the user's breathing pattern, namely, using the values of tidal volume and respiratory rate and the age of the user (Gilgen-Ammann, R., Koller, M., Huber, C. et al. Energy expenditure estimation from respiration variables. Sci Rep 7, 15995 (2017). https://doi.org/10.1038/s41598-017-16135-5).

The proposed method can be used to detect shortness of breath based on the actual breathing patterns of the user, and to provide recommendations for reducing the speed of movement or recommend special breathing exercises to compensate for shortness of breath. It is also possible to provide recommendations for rest based on the user's breathing patterns to restore the user's breathing, while recommending breathing exercises to restore breathing.

There is a range of safe parameters that the user should not exceed when performing physical exercises. For example, it is known that one should not breathe too deeply and sharply when there is a heavy load on the chest muscles. That is, if breathing patterns are detected that indicate that the user is breathing deeply and/or sharply, then a warning may be issued that with such a breathing pattern, one should not perform exercises with excessive load on the chest muscles.

During meditation, the device can monitor the depth of breathing and provide recommendations to the user to increase or decrease the depth of breathing.

If the breathing patterns reveal irregular heart rhythms, which in most cases corresponds to an increase in the user's stress level, breathing exercises aimed at reducing stress levels are recommended.

If the user smokes, the user may be given information on the inhalation/exhalation ratio and what effect this may have on the user's condition.

The authors conducted breathing studies involving 81 people. The studies were conducted both at rest and during physical activity. The results of the studies are shown in Fig. 7.

Fig. 7(A) shows the original PPG signal with weakly distinguishable fast oscillations of heart rate. The inset in Fig. 7(A) shows these oscillations on an enlarged scale. Fig. 7(B) shows the respiratory signal reconstructed from the PPG signal shown in the upper graph. The reconstructed respiratory signal shows oscillations at a significantly lower frequency compared to the heart rate, while the reconstructed respiratory signal preserves the shape of individual oscillations. This graph highlights two areas where the same user breathes differently: deep breathing and shallow breathing. The graph in Fig. 7(C) repeats the graph in Fig. 7(B), but it has marked extreme points responsible for the maxima and minima of the respiratory signal amplitude. The extreme points can be used to calculate the frequency and depth of breathing. Fig. 7(C) shows how the breathing patterns are located in the space of respiratory rate and depth of breathing (phase portraits). Accordingly, the asterisk is responsible for shallow breathing, the dot is for deep breathing, and the cross is for frequent deep breathing. In the study, no recommendations were given to users.

Although the invention has been described in connection with certain illustrative embodiments, it should be understood that the spirit of the invention is not limited to these specific embodiments. On the contrary, it is intended that the spirit of the invention include all alternatives, modifications and equivalents that may be included within the spirit and scope of the claims.

In addition, the invention retains all equivalents of the claimed invention, even if the claims of the invention are changed during the examination process.

References

1. E. Harbour, T. Stöggl, H. Schwameder and T. Finkenzeller, “Breath Tools: A Synthesis of Evidence-Based Breathing Strategies to Enhance Human Running,” Front. Physiol, v. 13, 2022|https://doi.org/10.3389/fphys.2022.813243 (https://www.frontiersin.org/articles/10.3389/fphys.2022.813243/full)

2. Breslav I.S. Breathing patterns: Physiology, extreme conditions, pathology - L.: "Nauka", 1984. - 206 p.

3. Borsch, M.K. External respiration patterns in the training system of highly qualified divers / M.K. Borsch, N.A. Paramonova, G.V. Popova // State and prospects of technical support of sports activities: collection of articles (procedures of the IV International scientific and technical conf.), Minsk, February 15-16, 2018. - Minsk: BNTU, 2018. - P. 82-88.

4. D.J. Meredith, D. Clifton, P. Charlton, J. Brooks, C. W. Pugh and L.Tarassenko, “Photoplethysmographic derivation of respiratory rate: a review of relevant physiology”, Journal of Medical Engineering & Technology, 2012; 1-7, Early Online.

5. Mitali R., Sapna Prabhu, “A Novel Algorithm to Obtain Respiratory Rate from the PPG Signal”, International Journal of Computer Applications 126(15):9-12, DOI:10.5120/ijca2015906263

6. R.S. Khandpur, Handbook of Biomedical Instrumentation, 2nd ed. New Delhi: Tata McGraw-Hill, 2010.

7. J.G. Webster, Medical Instrumentation: Application and Design, 3rd ed. New York: Wiley, 2007.

8. J. Prinable, P. Jones, D. Boland, A. McEwan, Cindy Thamrin, “Derivation of Respiratory Metrics in Health and Asthma,” Sensors 2020, 20(24), 7134; https://doi.org/10.3390/s20247134

Claims (131)

1. A breathing pattern monitoring device comprising: a memory unit capable of storing the user's personal data, as well as a database storing optimal breathing patterns recommended for each type of physical activity based on the user's personal data, a photoplethysmographic (PPG) sensor located on the user's body, configured to receive the user's PPG signal; an inertial measurement unit (IMU) sensor located on the user's body; a separation unit configured to extract from the PPG signal: - respiratory signal and - heart rate signal; a physical activity type recognition unit configured to recognize the user's physical activity type based on IMU sensor data and heart rate signal data; a classification unit configured to extract the actual breathing pattern from the respiratory signal, for which purpose the classification unit is configured to: a) arbitrarily set the duration of the time window, b) calculate the average value each respiratory parameter from the respiratory signal over a given time window; c) calculate the standard deviation from the average value ; d) compare standard deviation with an average value ; If for at least 70% of the breathing parameters, then analyze the breathing pattern with a given time window duration as the actual breathing pattern; If for at least 70% of the breathing parameters, then reduce the duration of the time window and perform steps b-d, If for at least 70% of the breathing parameters, then increase the duration of the time window and perform steps b-d; wherein the classification unit is configured to recognize the parameters of the actual breathing pattern from the actual breathing pattern; a comparison unit configured to compare the parameters of the actual breathing pattern with the parameters of the optimal breathing pattern recommended for the recognized physical activity, retrieved from the database, and also configured to detect a deviation of the parameters of the actual breathing pattern from the parameters of the optimal breathing pattern from the database, recommended for the recognized physical activity; a recommendation issuing unit configured to issue recommendations to the user on breathing technique based on the detected deviations of the parameters of the user's actual breathing pattern from the parameters of the optimal breathing pattern from the database recommended for the recognized physical activity. 2. The device according to claim 1, wherein the separation unit contains spectral filters. 3. The device according to paragraph 1, wherein the personal data contains height, weight, age, gender. 4. The device according to any one of paragraphs 1-3, wherein the breathing parameters are: volume of inhaled air, inhalation rate, exhalation time, volume of exhaled air, respiratory rate, time between exhalation and subsequent inhalation, duration of the breathing cycle, depth of breathing, duration of breath holding. 5. The device according to any one of paragraphs 1-4, wherein the recommendations are displayed on the screen of the user's wearable device. 6. A method for monitoring breathing patterns using a device according to claim 1, comprising the steps of: enter the user's personal data into the memory block; receive the user's PPG (photoplethysmogram) signal from the photoplethysmographic (PPG) sensor; receive data from the inertial measurement unit (IMU) sensor; the respiratory signal and the heart rate signal are separated from the PPG signal in a separation unit; recognize the type of physical activity of the user based on data from the IMU sensor and heart rate signal data in the physical activity type recognition unit; The actual breathing pattern is extracted from the respiratory signal and the parameters of the actual breathing pattern are extracted in the classification block, for this: a) arbitrarily set the duration of the time window, b) calculate the average value each respiratory parameter from the respiratory signal over a given time window; c) calculate the standard deviation from the average value ; d) compare the standard deviation with an average value ; If for at least 70% of the breathing parameters, the breathing pattern with a given time window duration is analyzed as the actual breathing pattern; If for at least 70% of the breathing parameters, then reduce the duration of the time window and perform steps b-d, If for at least 70% of the breathing parameters, then increase the duration of the time window and perform steps b-d; d) recognize breathing parameters from the actual breathing pattern; compare in the comparison block the parameters of the actual breathing pattern with the parameters of the optimal breathing pattern from the database recommended for the recognized physical activity; identify in the comparison block deviations of the parameters of the user's actual breathing pattern from the parameters of the optimal breathing pattern from the database recommended for the recognized physical activity; issue, via the issuing unit, recommendations to the user on breathing technique based on the detected deviations of the parameters of the user's actual breathing pattern from the parameters of the optimal breathing pattern from the database recommended for the recognized physical activity. 7. The method according to paragraph 6, in which, after entering personal data, a calibration measurement is additionally carried out to identify respiratory pathology, and in the event that respiratory pathology is identified, the user receives a recommendation to consult a doctor to compile an individual database. 8. A breathing pattern monitoring device comprising: a memory unit designed to store the user's personal data and a trained neural network; a photoplethysmographic (PPG) sensor configured to be positioned on the user's body and to receive the user's PPG signal; an inertial measurement unit (IMU) sensor configured to be positioned on the user's body; a separation unit configured to extract from the PPG signal: - respiratory signal and - heart rate signal; a physical activity type recognition unit configured to recognize the user's physical activity type based on IMU sensor data and heart rate signal data; a classification unit configured to extract the actual breathing pattern from the respiratory signal, for which purpose the classification unit is configured to: a) arbitrarily set the duration of the time window, b) calculate the average value each respiratory parameter from the respiratory signal over a given time window; c) calculate the standard deviation from the average value ; d) compare standard deviation with an average value ; If for at least 70% of the breathing parameters, then analyze the breathing pattern with a given time window duration as the actual breathing pattern; If for at least 70% of the breathing parameters, then reduce the duration of the time window and perform steps b-d, If for at least 70% of the breathing parameters, then increase the duration of the time window and perform steps b-d; a trained neural network capable of: - determining the optimal breathing pattern based on the user's personal data and the user's type of physical activity, - comparison of the parameters of the user's actual breathing pattern with the parameters of the said optimal breathing pattern; - obtaining an optimal respiratory signal based on comparison; a recommendation issuing unit configured to issue recommendations to the user on breathing technique based on the optimal respiratory signal issued by the neural network. 9. The device according to paragraph 8, wherein the personal data includes height, weight, age, gender. 10. The device according to any one of paragraphs. 8, 9, in which the neural network is additionally configured with the possibility determine whether the user is a generally healthy person or has a breathing pattern that is related to a pathological breathing pattern, If it is determined that the breathing pattern is related to a pathological breathing pattern, the issuing unit issues a recommendation to consult a doctor to compile an individual database of optimal breathing patterns corresponding to various types of physical activity. 11. A method for monitoring breathing patterns using a device according to claim 8, comprising the steps of: enter the user's personal data into the memory block; receive the user's PPG (photoplethysmogram) signal from the photoplethysmographic (PPG) sensor; receive data from the inertial measurement unit (IMU) sensor; via the separation block: extract the respiratory signal from the PPG signal, extract the heart rate signal; recognize the type of physical activity of the user based on the data IMU sensor and heart rate signal data via the physical activity recognition unit; extract the actual breathing pattern from the respiratory signal using a classification block, for this: a) arbitrarily set the duration of the time window, b) calculate the average value each respiratory parameter from the respiratory signal over a given time window; c) calculate the standard deviation from the average value ; d) compare the standard deviation with an average value ; If for at least 70% of the breathing parameters, the breathing pattern with a given time window duration is analyzed as the actual breathing pattern; If for at least 70% of the breathing parameters, then reduce the duration of the time window and perform steps b-d, If for at least 70% of the breathing parameters, then increase the duration of the time window and perform steps b-d; the user's personal data, the mentioned respiratory signal, the actual breathing pattern, the user's type of physical activity are fed to the input of the trained neural network, using a trained neural network: determine the optimal breathing pattern based on the user's personal data and the user's type of physical activity, compare the parameters of the actual breathing pattern of the user with the parameters of the said optimal breathing pattern, obtain the optimal respiratory signal based on the comparison; provide recommendations to the user on breathing technique, via a recommendation output unit, based on the optimal respiratory signal generated by the neural network. 12. A wearable device comprising a device according to claim 1, configured to implement a method for monitoring breathing patterns, comprising the steps of: enter the user's personal data into the memory block; receive the user's PPG (photoplethysmogram) signal from the photoplethysmographic (PPG) sensor; receive data from the inertial measurement unit (IMU) sensor; the respiratory signal and the heart rate signal are separated from the PPG signal in a separation unit; recognize the type of physical activity of the user based on data from the IMU sensor and heart rate signal data in the physical activity type recognition unit; The actual breathing pattern is extracted from the respiratory signal and the parameters of the actual breathing pattern are extracted in the classification block, for this: a) arbitrarily set the duration of the time window, b) calculate the average value each respiratory parameter from the respiratory signal over a given time window; c) calculate the standard deviation from the average value ; d) compare the standard deviation with an average value ; If for at least 70% of the breathing parameters, the breathing pattern with a given time window duration is analyzed as the actual breathing pattern; If for at least 70% of respiratory parameters, then reduce the duration of the time window and carry out stages b-g, If for at least 70% of the breathing parameters, then increase the duration of the time window and perform steps b-d; d) recognize breathing parameters from the actual breathing pattern; compare in the comparison block the parameters of the actual breathing pattern with the parameters of the optimal breathing pattern from the database recommended for the recognized physical activity; identify in the comparison block deviations of the parameters of the user's actual breathing pattern from the parameters of the optimal breathing pattern from the database recommended for the recognized physical activity; issue, via the issuing unit, recommendations to the user on breathing technique based on the detected deviations of the parameters of the user's actual breathing pattern from the parameters of the optimal breathing pattern from the database recommended for the recognized physical activity. 13. A wearable device comprising a device according to claim 8, configured to implement a method for monitoring breathing patterns, comprising the steps of: enter the user's personal data into the memory block; receive the user's PPG (photoplethysmogram) signal from the photoplethysmographic (PPG) sensor; receive data from the inertial measurement unit (IMU) sensor; via the separation block: extract the respiratory signal from the PPG signal, extracting a heart rate signal; recognizing the type of physical activity of the user based on the IMU sensor data and the heart rate signal data by means of a physical activity type recognition unit; extract the actual breathing pattern from the respiratory signal using a classification block, for this: a) arbitrarily set the duration of the time window, b) calculate the average value each respiratory parameter from the respiratory signal over a given time window; c) calculate the standard deviation from the average value ; d) compare the standard deviation with an average value ; If for at least 70% of respiratory parameters, then the breathing pattern with a given time window duration is analyzed as the actual breathing pattern; If for at least 70% of respiratory parameters, then reduce the duration of the time window and carry out stages b-g, If for at least 70% of respiratory parameters, then increase the duration of the time window and carry out stages b-g; the user's personal data, the mentioned respiratory signal, the actual breathing pattern, the user's type of physical activity are fed to the input of the trained neural network, using a trained neural network: determine the optimal breathing pattern based on the user's personal data and the user's type of physical activity, compare the parameters of the actual breathing pattern of the user with the parameters of the said optimal breathing pattern, obtain the optimal respiratory signal based on the comparison; provide recommendations to the user on breathing technique, via a recommendation output unit, based on the optimal respiratory signal generated by the neural network.